RCARS — RHDP Content Advisory & Recommendation System
AI-powered content intelligence for the Red Hat Demo Platform. RCARS reads every lab and demo in the RHDP catalog, understands what each one teaches, and uses that understanding to help teams find the right content, detect duplicate material, and identify items that should be retired.
What It Does
- Recommendations — Ask a question in plain English ("what should we show at a Kubernetes conference?") and get a ranked list of catalog items with rationales
- Content Overlap Detection — Pairwise semantic comparison across the catalog to surface duplicate or near-duplicate labs
- Retirement Analysis — Scores every catalog item based on provisions, sales impact, cost, and age to identify retirement candidates
- Infrastructure Metadata — Extracts workload mappings, cloud providers, and platform details from AgnosticD v2 configurations
- Catalog Browse — Filterable view of all catalog items with curator controls for content management
Architecture
Four deployments on OpenShift: React frontend, FastAPI API, arq workers (scan + recommend), PostgreSQL with pgvector. LLM analysis via LiteMaaS (primary) with Vertex AI fallback. Nightly pipeline handles catalog refresh, stale detection, workload scanning, content similarity, and reporting sync.
Documentation
Full documentation is published at rhpds.github.io/rcars.
Contributing
All changes must go through a pull request to the main branch. The project owner will review and merge. Direct pushes to main are blocked by branch protection.
Deployment
RCARS is deployed to OpenShift via Ansible. See ansible/ for playbooks and ansible/vars/common.yml for shared configuration. Environment-specific vars (dev.yml, prod.yml) are gitignored.
ansible-playbook ansible/deploy.yml -e env=dev --tags deploy
Local Development
./dev-services.sh start # PostgreSQL, Redis, API, workers, frontend
./dev-services.sh stop
See the Development guide for full setup instructions.
